提交 · 1219334d725df0d838d5e6801b65e338dede31a7 · testgroup / pytensor

28 11月, 2012 16 次提交
- Implemented UPLO parameter for eigh. · 1219334d
  由 abalkin 提交于 11月 12, 2012
  
  1219334d
- Simplified grad eigh calculations. · b9a6f0b7
  由 abalkin 提交于 11月 12, 2012
  
  b9a6f0b7
- Eliminated pinv from grad eigh calculations. · a00d8924
  由 abalkin 提交于 11月 12, 2012
  
  a00d8924
- Test grad wrt eigenvalues and eigenvectors separately. · 143103d3
  由 abalkin 提交于 11月 12, 2012
  
  143103d3
- Force symmetric matrix in eigh grad test. · 4d6fa38b
  由 abalkin 提交于 11月 12, 2012
  
  4d6fa38b
- Added a work-around for grad wrt disconnected variables. · cdb6f78f
  由 abalkin 提交于 11月 11, 2012
  
  cdb6f78f
- Stricter passing test. · 9df4514a
  由 abalkin 提交于 11月 11, 2012
  
  9df4514a
- Debugging math formulae. · c9c8017b
  由 abalkin 提交于 11月 10, 2012
  
  c9c8017b
- Split Eig and Eigh into different classes. · 25538faa
  由 abalkin 提交于 11月 10, 2012
  
  25538faa
- Implemented linalg.eigh. · 940bd5a8
  由 abalkin 提交于 11月 09, 2012
  
  940bd5a8
- Implemented Eig.grad(). Need to debug eigenvectors' component. · 7e3caf1e
  由 abalkin 提交于 11月 09, 2012
  
  7e3caf1e
- Removed redundant projection. · e4316b31
  由 abalkin 提交于 11月 08, 2012
  
  e4316b31
- More explanation of what Eig.grad() should return. · 47980073
  由 abalkin 提交于 11月 08, 2012
  
  47980073
- Better explanation of what Eig.grad() should return. · 8a51fa05
  由 abalkin 提交于 11月 08, 2012
  
  8a51fa05
- Added Eig.grad stub · 81b32a53
  由 abalkin 提交于 11月 08, 2012
  
  81b32a53
- Merge pull request #1088 from abalkin/issue-1080 · 8d7a8d56
  由 nouiz 提交于 11月 27, 2012
```
Issue #1080: Make TensorVariable interface more similar to that of numpy.ndarray
```
  8d7a8d56
27 11月, 2012 4 次提交
- Merge pull request #1097 from abalkin/doc-fixes · 1bb14956
  由 Olivier Delalleau 提交于 11月 27, 2012
```
Fixed a typo.
```
  1bb14956
- Fixed a typo. · ad24d986
  由 abalkin 提交于 11月 26, 2012
  
  ad24d986
- Merge pull request #1012 from goodfeli/cuda_alloc · 561b084b
  由 nouiz 提交于 11月 26, 2012
```
Ready to merge: Cuda allocation code
```
  561b084b
- Merge pull request #1094 from abalkin/doc-fixes · 9eb9db60
  由 nouiz 提交于 11月 26, 2012
```
Fixed a typo.
```
  9eb9db60
24 11月, 2012 5 次提交
- Fixed a typo. · af489ee9
  由 abalkin 提交于 11月 24, 2012
  
  af489ee9
- Merge pull request #1093 from goodfeli/name_fgraph · 5eeee66b
  由 nouiz 提交于 11月 23, 2012
```
make compiled function's fgraphs have same name as function
```
  5eeee66b
- make compiled function's fgraphs have same name as function · acd89e32
  由 Ian Goodfellow 提交于 11月 23, 2012
  
  acd89e32
- Merge pull request #1090 from nouiz/err_msg_debugmode · 9989c1e1
  由 lamblin 提交于 11月 23, 2012
```
Make work the new extra debug info just merged and add extra debug about...
```
  9989c1e1
- Make work the new extra debug info just merged and add extra debug about the output. · 254894fa
  由 Frederic 提交于 11月 23, 2012
```
This time I ran the code and saw the error to validate it. It is working correctly.
```
  254894fa
23 11月, 2012 15 次提交
- the TODO I completed was actually already done; updated the comments · 4557c310
  由 Ian Goodfellow 提交于 11月 23, 2012
```
to make it more clear how this function avoids unnecessary reallocation
```
  4557c310
- doc · 24890540
  由 Ian Goodfellow 提交于 11月 23, 2012
  
  24890540
- Reuse optimised special case code in diagonal(). · 2783b9cc
  由 abalkin 提交于 11月 22, 2012
  
  2783b9cc
- More tests for Diagonal. · ca068af0
  由 abalkin 提交于 11月 22, 2012
  
  ca068af0
- Merge pull request #1089 from lamblin/fix_softmaxgrad · 280f0c80
  由 nouiz 提交于 11月 22, 2012
```
Make opt work when shape has been specified.
```
  280f0c80
- Use debugprint instead of helper function. · 63c4209a
  由 Pascal Lamblin 提交于 11月 22, 2012
  
  63c4209a
- Make opt work when shape has been specified. · c880fff7
  由 Pascal Lamblin 提交于 11月 22, 2012
  
  c880fff7
- Deduplicate helper function. Update comment. · b624e3aa
  由 Pascal Lamblin 提交于 11月 22, 2012
  
  b624e3aa
- Merge remote-tracking branch 'upstream/master' into issue-1080 · 726aeae5
  由 abalkin 提交于 11月 22, 2012
  
  726aeae5
- WIP: Started implementation of diag(). · 55f0aa5d
  由 abalkin 提交于 11月 22, 2012
  
  55f0aa5d
- Return grad_not_implemented() from Diagonal.grad(). · 3de5ca32
  由 abalkin 提交于 11月 22, 2012
  
  3de5ca32
- WIP: Implemented non-zero offset case in Diagonal.infer_shape(). · 6e60fe28
  由 abalkin 提交于 11月 22, 2012
  
  6e60fe28
- WIP: Restricted offset and axes arguments to constants and implemented infer_shape. · 66ba51fe
  由 abalkin 提交于 11月 22, 2012
  
  66ba51fe
- WIP: Implemented diagonal(x, offset=0, axis1=0, axis2=1). · 6c67891a
  由 abalkin 提交于 11月 22, 2012
  
  6c67891a
- Merge pull request #1086 from nouiz/specify_shape · 7c6ebb2f
  由 goodfeli 提交于 11月 22, 2012
```
Add c_code to SpecifyShape and tests and assert in the op.
```
  7c6ebb2f